options.lua is a whole-file rewrite, so a caller handing saveOptions a
partial table (only the keys it changed) silently dropped every key it did
not mention: launcher-only keys like lastVersion, and keys the launcher set
(battleBg, tilt) all fell back to defaults.
saveOptions now reads the on-disk file first and folds caller-absent
values underneath before mergeOptions backfills defaults. A table holding
every defaultOptions key is a full snapshot and stays authoritative, so the
fold is inert for all in-repo writers (every one passes loadOptions-ed
tables) and cannot resurrect the bindings/activeProfile deletions the
RESET REBINDS and mod-manager paths make on full tables.
Adds a regression suite (options_partial_write_bug932.lua) pinning the
merge, and updates the #828 suite's partial-write assertion, which now
expects lastVersion to survive a delta write.
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
After the Fly departure animation finished (bird off-screen) and during
Dig/teleport, the trainer sprite popped back in standing at the old cell
for the whole 32-frame black fade-out before the transition. The
player-hide guard only held while a departure animation was live:
flyAnim went nil the instant path2 completed and the teleportOut
countdown cleared the spin fields at 0, but startWarpTo's Transition
(not isOpaque) keeps the overworld drawing beneath the veil, and the
arrival animation is not armed until setMap's midpoint.
Add a playerHidden flag on OverworldState that bridges the gap:
- set when each departure completes (flyAnim path2 / teleportOut hit 0),
immediately before the warp starts;
- cleared in startWarpTo's Transition enter callback, synchronously
after setMap and before the arrival arms flyArrive / spinDrop, so the
player is never drawable mid-fade and never bare on the landing frame;
- folded into both player-draw guards.
ROM-free regression test (tests/engine/warp_sprite_hidden_bug916.lua)
drives the REAL Transition + setMap headlessly for Dig and Fly and
asserts zero fade frames leave the player drawable bare (would have
observed 31/32 gap frames before the fix). Runs in the CI headless T2
tier.
Dig spin timing/lift and the black fade color are left as-is (fade is
intentional per #607).
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
trainers.battleTheme validated and merged onto the trainer record but was
never read: battle music came solely from data.audio.battle[kind] where
kind is computeMusicKind()'s final/gym/trainer/wild. Route both battle-
theme start sites through a single choke point:
- BattleState:playBattleTheme() cues Music.playBattle with the override
(self.trainer.battleTheme via battleTheme()), defaulting to the kind
when unset, so vanilla fights and #782's non-gym Giovanni are unchanged.
- BattleState:enter() and OverworldController:pushBattle() both call it.
- Music.playBattle gains an optional 4th song arg that overrides the kind
default, and real call sites now populate the music.select trainerId.
- Victory jingles stay kind-based: a custom battle theme has no derivable
win-variant.
New ROM-free T2 suite tests/engine/trainer_battle_theme_bug945.lua covers
mod load, override resolution, the choke point, and the nil-override
parity gate.
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
The mod update and Find Mods feeds feed the raw HTTP body straight to
Json.decode. When the endpoint hands back something that is not JSON
(an HTML error page, a proxy/captive prompt, or a plain-text outage
message like "Exceeded secondary rate limit" -- usually still HTTP 200),
the decoder's "unexpected character 'E'" assert escaped through the
pcall and became the error message, blaming the parser instead of the
response.
Add Json.describeUnexpected() as a pre-decode content-type guard: it
returns nil for body shapes the endpoints actually publish (JSON object
or array) and otherwise a short message naming what the server sent
(HTML page / plain text / empty, with a preview). Wire it into
ModUpdate.parseReleases and ModIndex.parse, so both the sync and async
update-check paths surface the real answer instead of the parse error.
HTTP status was already checked upstream by HostShell.httpGet (non-2xx
becomes "HTTP <code> from <url> (...)"); this closes the remaining
"2xx but not JSON" gap everywhere, including bridge platforms that
expose no status or headers.
Add regression tests for plain-text, HTML, and empty bodies; strengthen
the ModIndex HTML soft-fail test.
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Importing Red unpacked data/generated, assets/generated and
rom-cache.complete straight into the save-dir root, while Blue and
Yellow land under blue/ and yellow/. Red now uses cachePrefix red/
like the others.
CacheFs.migrateLegacyRedCache moves a pre-existing root cache into
red/ on first boot, from RomImporter.new before the readiness loop
and from mountVersion, so existing installs keep their import instead
of being asked for the ROM again. The move only runs when the root
marker resolves to the save directory, so a dev checkout's source
tree is never touched, and the portable game folder is skipped when
it is the physfs source.
Closes#899
The super-effective / not-very-effective hit sounds were only covered by
the manual ears-only driver (tests/drivers/hit_sfx_bug826_test.lua), so
the wFrequencyModifier wiring that fixed#826 (and that #902 reports as
a swap had no CI guard. This suite pins the polynomial-counter bytes
for all three hit sounds against the real audio/sfx programs: bare they
read swapped (super effective ends duller), and pitched they read
correct (super effective 4 shift 1 crack, not very effective
shift 10 thud). A port that drops the modifier fails 11 of 24 checks.
CLOSES#902
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
EOF
)
The modkit validate and pack drivers run the real loader under plain
luajit, with no love global. With --base imported, Data:load falls
back to CacheFs.readActive for a generated module require cannot find
(an optional module like data/generated/audio.lua is legitimately
absent from developer and stale caches), and CacheFs.read indexed
love.filesystem once there was no portable root, so validate and pack
died with MK100 before the mod was even looked at. Headless there is
no save directory to read from, so return nil like any other cache
miss.
Refs #850
The SAVE FILES card only accepted saves of exactly 32768 bytes and
refused anything else. importToSlot now classifies a non-32768 file by
the integrity of its main-data checksum instead:
- Oversize + valid checksum -> an emulator RTC footer, so the launcher
asks for confirmation, then truncates to 32768 on force.
- Oversize + invalid checksum -> rejected.
- Undersize + valid checksum -> imports zero-padded; otherwise refused.
Adds the "Oversized save file" confirm modal, a new vendor-oracle test
built by gen1lib (PKHeX-derived) run as its own Lua 5.4 tier, oversize/
truncated policy tests, and the LUA54 wiring in test.sh.
# Conflicts:
# src/import/LauncherView.lua
AverageConsumer
bryanthaboi
MaxTomahawk
Shane McGovern
Andrew Barnes
Caorthann
johnjohto
Andrew Quenehen
kikimanjaro